London AI Lab Inherent Raises $50m to Reinvent Science

Inherent, a London AI lab built by alumni of Google DeepMind, Microsoft and the Biden White House, has stepped out of stealth with a $50 million seed round and a goal that reads more like a manifesto than a product plan: rebuild the scientific method for an age of self-improving machines. The round, co-led by Index Ventures and Radical Ventures, was announced on May 28.

On its own, the cheque looks like one more European seed round. Placed next to the past year of dealmaking, it marks Europe’s late arrival into a venture category that barely existed two years ago and now carries billion-dollar price tags: the AI-native science lab.

A $50m Seed With a 400-Year Target

The pitch is deliberately grand. Inherent says it wants to “write the playbook for AI-native science,” and it has named its core system Faraday, after Michael Faraday, the 19th-century experimentalist who turned electromagnetism from a parlour curiosity into working engineering.

The mechanics of the deal are more concrete. Index Ventures and Radical Ventures co-led, with a supporting cast that includes NVentures (NVIDIA’s venture arm), Ex/Ante, Metaplanet, Macroscopic Ventures and Mythos Ventures. Law firm Wilson Sonsini advised the company on the transaction.

What buyers are not getting yet is a product. The details disclosed in the seed-round announcement describe a lab fresh out of stealth, with a research thesis and a founding team but nothing shipped to the public. That gap between ambition and output is exactly what makes the round worth reading closely.

The Wave Inherent Just Joined

Strip away the London address and Inherent fits a pattern that has built quietly over the past 18 months. A handful of labs have stopped framing AI as a tool that helps scientists and started selling the idea of the machine as the scientist, running hypotheses, designing experiments and learning from the results with limited human steering.

The numbers around those rivals dwarf Inherent’s seed. Lila Sciences, based in Cambridge, Massachusetts, has raised roughly $550 million and was valued near $1.3 billion after fresh backing late last year, according to its PitchBook funding profile. Periodic Labs, founded by former OpenAI and DeepMind researchers Liam Fedus and Ekin Dogus Cubuk, banked a $300 million seed and, per Bloomberg reporting in March, was negotiating a much larger round at an implied valuation near $7 billion.

Money is also flowing from the other side of the table. GSK committed $50 million up front to NOETIK in early 2026 for access to its oncology foundation model, one of several pharmaceutical deals that treat AI discovery platforms as strategic infrastructure rather than experiments. The backdrop is a venture market that has tilted hard toward the sector, with AI absorbing close to 80% of all startup funding in the first quarter of the year.

Set against that table, Inherent is the smallest and youngest name, and the only European one.

What Faraday Is Built to Do

Inherent describes its system as a way for humans and self-improving AI to work together on the hardest problems in science. The word that keeps surfacing in the company’s framing is recursive: the research improves the organisation, and the organisation reshapes the research, in a continuous loop.

Danny Rimer, the Index Ventures partner backing the deal, has described the goal as a reimagining of the scientific method from first principles, rather than AI bolted onto methods scientists have used for 400 years. His firm’s investment note pushes the point further, arguing that most AI models stumble at genuine discovery because they are trained to give good answers, not to work out which questions are worth asking.

That distinction is the whole bet. Today’s large models excel at compressing what is already known; discovery requires generating and testing ideas that are not yet in any training set.

Index frames the ambition in unusually sweeping terms.

AI-native science will look and feel totally different to the scientific method we’ve grown used to over the past 400 years.

You can read the full investment thesis behind the Faraday system on the investor’s site, where the partners compare a future lab to something a medieval monk would find alien. It is a vision long on metaphor and, for now, short on demonstrated results.

From DeepMind and the White House to a London Lab

The reason serious money showed up at seed stage is the team. Three of the four co-founders come straight out of Google DeepMind, and one carries policy credentials that are rare in a research startup.

• Tantum Collins, a former DeepMind researcher who also worked on AI policy at the Biden White House.
• Edward Hughes, who comes from DeepMind’s research ranks.
• Louis Kirsch, a DeepMind alumnus focused on machine-learning research.
• Kaloyan Aleksiev, who built infrastructure at Reka AI and Microsoft.

The lab also leans on Matt Clifford, co-founder of talent investor Entrepreneurs First and a former government AI adviser, who sits as an adviser to Inherent and called the founders some of the “most impressive, thoughtful founders” he had met. In a London scene that has produced a steady run of well-funded AI startups, from OpenAI and Spotify leaders backing a London AI security startup to a new wave of model labs, that kind of pedigree opens doors fast.

Why NVIDIA Keeps Showing Up on the Cap Table

One name on Inherent’s investor list deserves a second look. NVentures, the venture arm of NVIDIA, joined the round, and it has also backed Lila Sciences. The chipmaker is quietly seeding the labs most likely to become heavy buyers of its hardware.

It is a classic picks-and-shovels position. AI-native science is compute-hungry by design, since training models to run thousands of simulated experiments and self-improvement loops burns through processing power at a scale ordinary software never approaches. You can read NVIDIA’s own framing on its corporate venture investing page, which leans toward startups that expand demand for accelerated computing.

For Inherent, the backing is validation and a hedge at once. It signals that a strategic investor sees the thesis, and it ties the lab’s compute future to a supplier that now has a financial stake in its success.

Seed Money Against Billion-Dollar Rivals

The risks sit in plain sight. Inherent is entering at seed stage against competitors already valued in the billions, with deeper benches, physical lab capacity and, in some cases, a year or more of head start. A $50 million cheque is generous for a London seed and modest for a frontier AI ambition.

Then there is the question of returns. Investors who have chased AI materials and discovery platforms have learned that turning a clever model into a marketable breakthrough is a long, capital-intensive road, and that lab demonstrations rarely survive contact with real-world manufacturing on the first try. A pitch built on reimagining the scientific method is harder to falsify, and harder to price, than a drug candidate or a battery chemistry.

The wider European funding climate at least gives Inherent cover. Capital has been flowing into the region’s AI names at a pace not seen before, from ElevenLabs securing a $500m round amid a European tech funding surge to a string of model and infrastructure deals. That tide makes it easier to raise the much larger sums Inherent will need if Faraday is to graduate from thesis to working system.

If the lab produces a genuine discovery in the next two years, the seed price will look like a bargain and London will own a flagship in a field the United States currently dominates. If it stays a manifesto, Inherent becomes a cautionary line in the story of how much investors paid to find out whether machines can do science on their own.